.png){kind=icon}
.png){kind=icon}
Return
A High-Accuracy Deep Learning Pipeline for Orange Leaf Disease Classification Using YOLOv10 Detection and MobileNetV3-Small
.png){kind=icon}
PRE
AI
P
H
Q
DOI：10.1142/S1469026826500094.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This study presents a comprehensive technological framework for detecting and classifying citrus leaf diseases using advanced deep learning convolutional neural networks (CNNs). We propose an innovative approach that integrates YOLOv10 for object detection and MobileNetV3 for disease classification, referred to as YOLO10.MNET3, to address key challenges in precision agriculture and plant pathology. The YOLOv10 object detection model demonstrates superior performance through advanced architectural design and novel training strategies, including consistent dual labeling and the elimination of traditional nonmaximum suppression (NMS) algorithms. These optimizations significantly improve inference efficiency and detection accuracy. To ensure precise disease classification within the detected leaf regions, we deploy MobileNetV3, a lightweight neural network specifically designed for resource-constrained environments. For real-world implementation, we convert the YOLOv10 model to NCNN, a high-performance neural network framework that significantly reduces model complexity and inference latency. The proposed method successfully identifies and classifies five distinct citrus leaf conditions: Citrus canker, Citrus greening, Citrus mealybugs, Spiny whitefly, and Healthy, achieving an impressive overall accuracy of 99.7%. Experimental results validate the robustness and effectiveness of the proposed deep learning approach, demonstrating its potential for real-time crop disease monitoring in precision agriculture and computationally limited environments.
Keywords:
Citrus classification
deep learning
image recognition
ONNX
smart agriculture
model optimization
quality management
dataset
Journal
I
IF:
1.3
Papers:
24
Citations:
0
Organization
Cited Papers